Labom – Temperature measurement in a regulated environment
4 July 2016
Measuring the medium temperature is a common task in the pharmaceutical industry. This is usually achieved with invasive measuring systems which reach into the process area. This procedure is common and robust. The downsides of this approach are that it is not suited for small pipe diameters and that the necessary process interfaces add an additional hygienic risk. An alternative way to derive the medium temperature is by measuring the pipe surface temperature. This is a detailed description of the strengths and limitations of this measurement procedure. It shows under which process and ambient conditions measurement of the pipe surface can be a flexible and cost-effective alternative to established methods.
Temperature measurements are the most common measurement tasks in pharmaceutical plants. In addition to monitoring and control of the production process, measurement and documentation of the achieved steam temperature during the customary plant sterilisation with saturated steam (SIP) are also required. Due to the need to process the measurement result and meet the documentation requirements, almost all measuring devices used are electronic. Therefore, this paper does not consider mechanical dial gauges.
Advantages and disadvantages of the invasive temperature measurement method
Fig. 1: Examples for invasive temperature measuring devices (Source for all images: LABOM).
The most common type of temperature measurement uses a temperature-sensitive sensor, often a Pt100 element, which is inserted into the process. Use of such devices is referred to as „invasive measurement“ below, since it „invades“ the process area. Various models (Fig. 1) of this type of measuring device by various manufacturers are available in the market. Clamp connections and aseptic screw connections (e.g., as per DIN 11864) are usually used in the pharmaceutical industry. The measuring device is usually designed in such a way that it can be removed easily for the required calibration without interrupting the process, e.g., by using a replaceable measurement probe. Varying the sensor length allows for very precise adjustment of the measurement position. In summary, this measurement procedure is very stable and well-tested in practice.
Despite all these advantages, the invasive measurement method also has some serious disadvantages. Pharmaceutical products are often produced in small batches; therefore, the used pipes also have very small diameters. A sensor tip reaching into the process area can be a significant obstacle to the flow. The minimum pipe diameter for this device class is DN10, which is also the minimum size of the widely used Varivent® process connections. The interfaces between the process connection and the system and between the measuring device and the process connection also present a significant hygienic risk. The maintenance requirements are also higher, since the implemented elastomer seals usually need to be replaced regularly. Therefore, the total costs over the lifetime of the measuring point significantly exceed the price of the device.
Surface temperature measurement method
Fig. 2: Example models for pipe surface measurement devices.
To avoid the mentioned problems of invasive temperature measurement, the pipe surface measurement technique – as already widely used in heating systems – has been applied to pharmaceutical measurement tasks. Measurement of the pipe surface avoids any interference with the process. This helps preclude hygiene risks caused by the measurement, and very small pipe diameters are no longer a problem. The devices are usually quick and easy to install and can be integrated into existing systems very easily. This might be required if, during the final approval inspection or at the commissioning of the system, it becomes apparent that the available temperature measurements are inadequate. Two typical models are shown as examples in Fig. 2. In the first example, a contact element establishes the connection to the pipe. The measurement probe is then inserted into the contact element. The advantage of this method is that the same type of measurement probe as for invasive measurement can be used, therefore the calibration procedure is also the same. However, this version does not always fulfil the requirements for response time and measurement accuracy. The second example shows a version where the measurement probe has direct contact to the pipe surface. This can significantly reduce the heat capacity of the measuring device, and hence the response time. The measurement accuracy is also higher than for the first version. However, the surface temperature measurement method is not without disadvantages. The achievable accuracy depends significantly on the process conditions, as will be shown in detail below. Installation errors can also have an influence on the measurement. The achievable measurement accuracy is also viewed critically because there is little available experience with this relatively new technology. Therefore, we take a theoretical look at the measurement of the pipe surface temperature below, in order to derive suggestions for practical use.
Influence of the process conditions on the achievable accuracy
To achieve an acceptable measurement accuracy, the pipe surface must have nearly the same temperature as the medium. Due to the laws of thermodynamics, the pipe surface can never have exactly the same temperature as the medium. This means that it is necessary to judge whether the occurring deviation is small enough. This also applies for invasive measurement. However, experience in practice has already shown that the measurement deviation is usually within acceptable limits. Apart from the temperature of the medium, the pipe surface temperature depends on three thermal resistances:
1. The thermal resistance between the pipe centre and the inner wall of the pipe
2. The thermal resistance of the pipe
3. The thermal resistance between the outer wall of the pipe and the surroundings
Similar to the way in which electric resistance decreases voltage, thermal resistance decreases the temperature. This is shown by the red curve in Fig. 3. In the ideal case, the entire temperature drop results only from the thermal resistance between the pipe surface and the surroundings. In this case, the temperature of the
Fig. 3: Diagram of temperature curve across a pipe.
medium and of the pipe surface would be identical. For purposes of technical measurement, this means that the sum of the first two thermal resistances must be very small compared to the resistance between the pipe surface and the surroundings. If that is the case, the pipe surface temperature would be close to the temperature of the medium.
Theoretical analysis of influencing factors
For a quantitative analysis, all three aforementioned thermal resistances must be considered simultaneously, since each one influences the occurring flow of heat and hence the temperatures at the transition surfaces. Unless stated otherwise, the quantitative calculations are based on a stainless steel pipe with measurements of 13.5 x 1.6 mm in still ambient air with a temperature of 20 °C. These are the usual conditions for pharmaceutical systems in closed buildings. For the underlying formulas, refer to the relevant literature, such as the VDI Heat Atlas .
Transfer of heat from the medium to the pipe
Fig. 4: Comparison of temperature profiles for laminar and turbulent pipe flows.
The thermodynamic process during the transfer of heat from the medium to the pipe is called „forced convection“. The pressure difference which occurs during the process forces the medium to move through the pipe. One crucial factor is the type of flow which occurs. We distinguish between laminar and turbulent flow (see Fig. 4). In case of laminar flow, there is no whirling and the particles of the medium only move in the flow direction. Therefore, the only mechanism for the transfer of heat to the inner wall of the pipe is the heat conduction within the medium. At the same time, friction at the pipe wall slows down the outer layers of the medium, which gives it more time to cool down. Laminar flow therefore leads to large differences in the distribution of speed and temperature within the pipe. In the case of turbulent flow, the particles of the medium move without any particular order. Therefore, heat transportation occurs in addition to heat conduction. Whirls move hot particles from the pipe centre closer to the wall of the pipe and thus transport heat energy. As a result, there is a largely homogeneous distribution of speed and temperature. There is only a small temperature difference in the vicinity of the pipe wall due to friction. It is easy to imagine that turbulent flow is much better suited for measuring the temperature at the pipe surface, since, in laminar flow, the inner temperature of the pipe already deviates significantly from the temperature of the medium. Which type of flow occurs depends on the flow speed, the pipe geometry and on the properties of the medium. The most relevant parameter is the medium‘s viscosity.
The higher the viscosity, the more likely there will be a laminar flow. For calculation purposes, the pipe wall thickness is set to 0 mm, to exclude the pipe as an influencing factor.
|Temperature drop inside pipe depending on medium.|
|Water||50 °C 2m/s||0.10 K|
|Methanol||50 °C 2m/s||0.25 K|
|Glycerine||50 °C 2m/s||3.99 K|
|Air||120 °C 20m/s||12.8 K|
|Saturated steam||120 °C 20m/s||0.11 K|
The results of the calculation are shown in Table 1. For water, there is a temperature difference of only 0.10 K between the medium and the inner wall of the pipe. The very good heat conductivity of water has an influence here. For methanol, the deviation amounts to 0.25 K under the same conditions. Despite its lower viscosity compared to water, the influence of methanol‘s poor heat conductivity is stronger. The high viscosity of glycerine leads to a laminar flow, which results in a large temperature difference (as explained above) of almost 4 K. For air, there is a very significant deviation of 12.8 K. This is due to the fact that gases in general have a very low heat capacity. The emission of even a comparatively small amount of heat energy leads to a strong temperature decrease. For saturated water steam, as commonly used in SIP processes, the deviation is only 0.11 K, although steam is a gas. This is due to the special influence of condensation. The hot steam condenses on the inner surface of the pipe and passes on its condensation heat directly to the pipe. All in all, we can say that surface temperature measurement is particularly useful for water and watery solutions, as well as saturated steam. Alcohols are slightly less suitable. For gases and viscous mediums, such as oils, this method is only useful to measure trends or requires additional insulation (see below).
Heat conduction inside the pipe
The transfer of heat in a pipe results from heat conduction and depends on only three factors:
• Pipe diameter
• Pipe wall thickness
• Heat conductivity of the pipe material
The heat conductivity of technical materials ranges across four decimal powers. At the upper end, we have metals such as silver, copper and aluminium, whose conductivity is between 300 to 400 W/mK; at the lower end, we have technical insulators and air (0.02 to 0.03 W/mK). Stainless steel, a material frequently used in the pharmaceutical industry, only has 15 W/mK and is hence one of the least conductive metallic materials. For the calculations (also see Table 2), a homogeneous medium temperature of 120 °C is assumed.
|Temperature drop inside pipe depending on
material and geometry.
|Stainless Steel 13.5 x 1.6 mm||0.13 K|
|Stainless Steel 27.0 x 1.6 mm||0.11 K|
|Stainless Steel 13.5 x 3.2 mm||0.32 K|
|Plastic (PC) 13.5 x 1.6 mm||8.79 K|
Under these conditions, the temperature difference between the inner and outer sides of the pipe is only 0.13 K. Doubling the diameter only slightly affects this value. Doubling the wall thickness leads to a deviation of 0.32 K. The wall thickness therefore has a significant effect on the temperature measurable on the outside of the pipe. However, the deviation is only a few tenths of a Kelvin, even when unusually thick walls are used. For a plastic pipe made of polycarbonate, the significantly lower heat conductivity of the material leads to an unacceptable deviation of 8.79 K. In summary, we can state that surface measurement requires a metallic pipe.
Transfer of heat from the pipe to the surroundings
The thermodynamic process for the transfer of heat from the pipe to still ambient air is called „free convection“. The ambient air heats up, becomes less dense and rises. Heat is transported away; hence there is a stronger heat flow than in the case of heat conduction only. If the air moves, e.g., due to wind in outdoor applications, free and forced convection overlap. For calculation purposes, we assume that the medium is saturated steam with a temperature of 120 °C, since this is a typical case for the sterilisation of pharmaceutical systems.
|Temperature difference between medium/outer pipe wall, depending on surroundings.|
|Pipe without insulation||With 1cm of PU foam insulation
l = 0.03 W/(mK)
|With 1cm plastic jacket (PVDF)
l = 0.17 W/(mK)
|Still air||0.28 K||0.09 K||0.29 K|
|Moving air (1m/s)||0.87 K||0.11 K||0.48 K|
(5m/s -3 BFT)
|2.45 K||0.12 K||0.61 K|
The overview in Table 3 differentiates between still air (indoor applications), slight air movement (1 m/s; e.g., due to nearby AC or vent) and strong air movement (5 m/s; outdoor applications). We also look at the influence of additional insulation. For an uninsulated pipe in still air, the most common application in pharmaceutics, the total temperature difference between the medium inside the pipe and the outer surface of the pipe is only 0.28 K. Without insulation, however, this value is significantly worse in case of moving air. However, PU foam insulation with a thickness of only one centimetre almost completely eliminates the influence of air movements. The temperature deviation in still air is also reduced by two thirds. Where there is a risk of air movement or if high accuracy is required, insulation can thus lead to a significant improvement. A plastic jacket (1 cm of PVDF with λ = 0.17 W/mK in this case) also reduces the susceptibility to air movements. In still air, however, there is a slight deterioration, which is interesting. The jacket increases the surface, so that more heat is passed on to the ambient air. This effect can be stronger than the insulating effect of the jacket, as is the case here. This effect is used for electrical cables, for example. Here, insulation lets the cables lose more heat; therefore, they can transport more electricity than an uninsulated wire. In summary, we can say that the deviation amounts to a few tenths of a Kelvin for applications in still air; in the case of strong air movement, similar results can be achieved by adding insulation.
Influence of the meter on the measurement
The aforementioned calculations are not sufficient to make a statement about the achieved accuracy of measurement. The measurement methods used for industrial applications are largely contacting procedures, e.g., with a Pt100 element. This leads to additional heat transfers, which influence the total flow of heat and hence the temperature at the sensor element. The following factors are to be considered:
• The transfer of heat from the pipe to the device
• The transfer of heat from the contact point to the actual sensor element (e.g., a Pt100 resistor)
• The transfer of heat from the sensor element to the device surface
• The transfer of heat from the device surface to the surroundings.
As was the case in the analysis of the pipe, the thermal resistance all the way to the sensor element must be very small compared to the thermal resistance between the sensor element and the surroundings. As the device surface is necessarily larger than the enclosed pipe surface, the effect described above occurs. The greater surface leads to heat being transferred to the surroundings more easily than by the pipe alone. As a result, the deviations calculated in theory cannot be achieved in practice. Design requirements on devices for measuring the pipe surface temperature Due to the situation mentioned above, the device has to be designed in a way which accommodates these circumstances and thus minimizes error. One option is to integrate insulating elements into the enclosure design. Integrated air chambers, for example, help take advantage of the very low heat conductivity of air. In addition, the sensor element and its support must be thermally insulated as much as possible from the rest of the enclosure and must have a low thermal mass in order to achieve fast response times. For consistently stable measurement, the thermal expansion of various materials must also be compensated. Especially in the case of large temperature fluctuations in steam-sterilised systems, the contact between the measuring device and the pipe can loosen over time, which can falsify the measuring result. The large number of possible device models and influencing factors makes it very difficult to analytically calculate the expected measuring deviation. Tests have shown, however, that a deviation of only one Kelvin is possible for steam sterilization applications in pharmaceutical systems. Quite interestingly, where the response time to abrupt temperature changes is concerned, the advantage can be on the side of pipe surface measurement, if an appropriate design is used. The mass to be heated is particularly important for this process. In particular in cases where thermowells are used, invasive devices are at a disadvantage.
Fig. 5 summarizes the results of the analysis. The invasive measurement method is tried and tested and can be used universally, as long as the pipe diameter is not too small. The surface measurement method is more sensitive to process and ambient conditions. It is particularly useful for measuring water and watery solutions, as well as saturated steam in indoor applications. For measurement in other mediums, we expect a higher deviation, but insulation can be used to reduce it. Especially in pharmaceutical systems, there are many temperature measurement points which are suitable for measuring the surface temperature. These systems are usually operated in closed rooms. Simultaneously, there is a trend towards smaller systems with smaller pipe diameters. The common method of steam sterilisation requires the sterilisation temperature to be monitored at various locations in the system. Here the advantages of surface temperature measurement, i.e. lower system costs and more flexibility, can be fully utilized.
Literature  VDI Heat Atlas, Springer Verlag 2006, ISBN 978-3-540-77876-9 Author Dr.-Ing. Thomas Köster Thomas Köster holds a doctorate in engineering. He studied precision engineering, measurement and control systems at the Technische Universität Braunschweig and completed his doctorate at the Leibniz Universität Hannover. He then spent several years developing complex mechatronic devices and systems. He has been R & D Manager at LABOM GmbH since 2009 and is responsible for all technical aspects of the product portfolio.